Agentic Commerce World: An Auditable and Verifiable Environment for Vibe Commerce
ACWorld tests buyer and merchant agents in shared transactions, with validation and logs built into the market.
The paper frames “vibe commerce” as natural-language buying or selling delegated to AI agents. Its Vibe Commerce Protocol checks agent actions before transaction state changes, then records the interaction for auditability and reproducible evaluation. The benchmark includes 200 capability-coverage tasks and 60 large-catalog tasks over 785,022 listings. Across ten models, reported mean scores span 65.9% to 85.6% on one track and 56.1% to 91.4% on the other. ArXiv · AI/CL/LG's note
The paper frames “vibe commerce” as natural-language buying or selling delegated to AI agents. Its Vibe Commerce Protocol checks agent actions before transaction state changes, then records the interaction for auditability and reproducible evaluation. The benchmark includes 200 capability-coverage tasks and 60 large-catalog tasks over 785,022 listings. Across ten models, reported mean scores span 65.9% to 85.6% on one track and 56.1% to 91.4% on the other. ArXiv · AI/CL/LG's note
score 4